|
|
WAsP – the Wind Atlas Analysis and Application Program |
|
|
WAsP – the Wind Atlas Analysis and Application Program |
Originally WEng applied a method developed by Leif Kristensen et al (2000) J. Wind Eng. Ind. Aerodyn. 87 147-166. However, we learned that sometimes it did not perform well, especially when the sector-wise Gumbel slopes differed a lot. To avoid such problems, WEng now stores sets of annual extremes at the reference site of the OEWC file or the standard conditions of the REWC file. You can see this if you open these files with an XML editor. For predictions we first use the flow model of WEng to calculate individual speedup and deflection for all sectors and all turbine sites, then we scale the annual extreme from the OEWC/REWC file, then we fit Gumbel distributions to the modelled local extreme winds, and finally we extrapolate to fifty-year extreme winds. This is done for each turbine site and each wind sector.
WEng applies the Gumble distribution but the litterature offers alternatives. Of particular interest is the generalized extreme distribution, which has the Gumble distribution as a special case. According the Gumble distribution the annual maxima falls close to a straight line in the probability plot used in WEng and WACA. With the generalized extreme-wind distribution, which has three model parameters instead of two, the theoretical distribution would appear to curve. To prove that the generalized extreme distribution is a significantly better model than the Gumbel distribution the observations should be clearly outside the error range shown by two grey curves in WEng. The natural scatter induced by the limited amount of observations available for typical wind energy studies, makes it difficult to distinguish between the models.
Alternative methods for fitting extreme-wind distributions fall into two major categories: periodic-maxima and peak-over-threshold methods. WEng and WACA apply an annual maximum method, fitted by probability-weighted moments, see Abild et al (1992) J. Wind Eng. Ind. Aerodyn. , 41 473-484. Comparison studies in the literature suggest that different methods are in general agreement, although some methods may be preferred due to the statistical performance. The annual maximum analysis of WEng and WACA is judged to be robust, but maybe slighlty conservative. The main reason why we do not offer alternatives is that WEng must convert each extreme-wind observation to site conditions before the extreme-wind statistics can be modeled at individual turbine sites. So, with peak-over-threshold methods, WEng would have to compute more flow situations than with the present method.
Extreme wind analysis based on site observations are difficult in areas with tropical cyclones. The high-wind area near the 'wall of the eye' is relatively limited and may not hit a particular site for several years so there is a risk that even 20 years of measurements may not be representative. See the Risø-R-1544 report for an alternative method based on 'best-track estimates' from meteorological institutes.
